ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent-MP/MP.pm
Revision: 1.131
Committed: Fri Mar 9 19:07:53 2012 UTC (12 years, 2 months ago) by root
Branch: MAIN
Changes since 1.130: +10 -5 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 =head1 NAME
2
3 AnyEvent::MP - erlang-style multi-processing/message-passing framework
4
5 =head1 SYNOPSIS
6
7 use AnyEvent::MP;
8
9 $NODE # contains this node's node ID
10 NODE # returns this node's node ID
11
12 $SELF # receiving/own port id in rcv callbacks
13
14 # initialise the node so it can send/receive messages
15 configure;
16
17 # ports are message destinations
18
19 # sending messages
20 snd $port, type => data...;
21 snd $port, @msg;
22 snd @msg_with_first_element_being_a_port;
23
24 # creating/using ports, the simple way
25 my $simple_port = port { my @msg = @_ };
26
27 # creating/using ports, tagged message matching
28 my $port = port;
29 rcv $port, ping => sub { snd $_[0], "pong" };
30 rcv $port, pong => sub { warn "pong received\n" };
31
32 # create a port on another node
33 my $port = spawn $node, $initfunc, @initdata;
34
35 # destroy a port again
36 kil $port; # "normal" kill
37 kil $port, my_error => "everything is broken"; # error kill
38
39 # monitoring
40 mon $port, $cb->(@msg) # callback is invoked on death
41 mon $port, $localport # kill localport on abnormal death
42 mon $port, $localport, @msg # send message on death
43
44 # temporarily execute code in port context
45 peval $port, sub { die "kill the port!" };
46
47 # execute callbacks in $SELF port context
48 my $timer = AE::timer 1, 0, psub {
49 die "kill the port, delayed";
50 };
51
52 =head1 CURRENT STATUS
53
54 bin/aemp - stable.
55 AnyEvent::MP - stable API, should work.
56 AnyEvent::MP::Intro - explains most concepts.
57 AnyEvent::MP::Kernel - mostly stable API.
58 AnyEvent::MP::Global - stable API.
59
60 =head1 DESCRIPTION
61
62 This module (-family) implements a simple message passing framework.
63
64 Despite its simplicity, you can securely message other processes running
65 on the same or other hosts, and you can supervise entities remotely.
66
67 For an introduction to this module family, see the L<AnyEvent::MP::Intro>
68 manual page and the examples under F<eg/>.
69
70 =head1 CONCEPTS
71
72 =over 4
73
74 =item port
75
76 Not to be confused with a TCP port, a "port" is something you can send
77 messages to (with the C<snd> function).
78
79 Ports allow you to register C<rcv> handlers that can match all or just
80 some messages. Messages send to ports will not be queued, regardless of
81 anything was listening for them or not.
82
83 Ports are represented by (printable) strings called "port IDs".
84
85 =item port ID - C<nodeid#portname>
86
87 A port ID is the concatenation of a node ID, a hash-mark (C<#>)
88 as separator, and a port name (a printable string of unspecified
89 format created by AnyEvent::MP).
90
91 =item node
92
93 A node is a single process containing at least one port - the node port,
94 which enables nodes to manage each other remotely, and to create new
95 ports.
96
97 Nodes are either public (have one or more listening ports) or private
98 (no listening ports). Private nodes cannot talk to other private nodes
99 currently, but all nodes can talk to public nodes.
100
101 Nodes is represented by (printable) strings called "node IDs".
102
103 =item node ID - C<[A-Za-z0-9_\-.:]*>
104
105 A node ID is a string that uniquely identifies the node within a
106 network. Depending on the configuration used, node IDs can look like a
107 hostname, a hostname and a port, or a random string. AnyEvent::MP itself
108 doesn't interpret node IDs in any way except to uniquely identify a node.
109
110 =item binds - C<ip:port>
111
112 Nodes can only talk to each other by creating some kind of connection to
113 each other. To do this, nodes should listen on one or more local transport
114 endpoints - binds.
115
116 Currently, only standard C<ip:port> specifications can be used, which
117 specify TCP ports to listen on. So a bind is basically just a tcp socket
118 in listening mode thta accepts conenctions form other nodes.
119
120 =item seed nodes
121
122 When a node starts, it knows nothing about the network it is in - it
123 needs to connect to at least one other node that is already in the
124 network. These other nodes are called "seed nodes".
125
126 Seed nodes themselves are not special - they are seed nodes only because
127 some other node I<uses> them as such, but any node can be used as seed
128 node for other nodes, and eahc node cna use a different set of seed nodes.
129
130 In addition to discovering the network, seed nodes are also used to
131 maintain the network - all nodes using the same seed node form are part of
132 the same network. If a network is split into multiple subnets because e.g.
133 the network link between the parts goes down, then using the same seed
134 nodes for all nodes ensures that eventually the subnets get merged again.
135
136 Seed nodes are expected to be long-running, and at least one seed node
137 should always be available. They should also be relatively responsive - a
138 seed node that blocks for long periods will slow down everybody else.
139
140 For small networks, it's best if every node uses the same set of seed
141 nodes. For large networks, it can be useful to specify "regional" seed
142 nodes for most nodes in an area, and use all seed nodes as seed nodes for
143 each other. What's important is that all seed nodes connections form a
144 complete graph, so that the network cannot split into separate subnets
145 forever.
146
147 Seed nodes are represented by seed IDs.
148
149 =item seed IDs - C<host:port>
150
151 Seed IDs are transport endpoint(s) (usually a hostname/IP address and a
152 TCP port) of nodes that should be used as seed nodes.
153
154 =item global nodes
155
156 An AEMP network needs a discovery service - nodes need to know how to
157 connect to other nodes they only know by name. In addition, AEMP offers a
158 distributed "group database", which maps group names to a list of strings
159 - for example, to register worker ports.
160
161 A network needs at least one global node to work, and allows every node to
162 be a global node.
163
164 Any node that loads the L<AnyEvent::MP::Global> module becomes a global
165 node and tries to keep connections to all other nodes. So while it can
166 make sense to make every node "global" in small networks, it usually makes
167 sense to only make seed nodes into global nodes in large networks (nodes
168 keep connections to seed nodes and global nodes, so makign them the same
169 reduces overhead).
170
171 =back
172
173 =head1 VARIABLES/FUNCTIONS
174
175 =over 4
176
177 =cut
178
179 package AnyEvent::MP;
180
181 use AnyEvent::MP::Config ();
182 use AnyEvent::MP::Kernel;
183 use AnyEvent::MP::Kernel qw(%NODE %PORT %PORT_DATA $UNIQ $RUNIQ $ID);
184
185 use common::sense;
186
187 use Carp ();
188
189 use AE ();
190 use Guard ();
191
192 use base "Exporter";
193
194 our $VERSION = $AnyEvent::MP::Config::VERSION;
195
196 our @EXPORT = qw(
197 NODE $NODE *SELF node_of after
198 configure
199 snd rcv mon mon_guard kil psub peval spawn cal
200 port
201 db_set db_del db_reg
202 db_mon db_family db_keys db_values
203 );
204
205 our $SELF;
206
207 sub _self_die() {
208 my $msg = $@;
209 $msg =~ s/\n+$// unless ref $msg;
210 kil $SELF, die => $msg;
211 }
212
213 =item $thisnode = NODE / $NODE
214
215 The C<NODE> function returns, and the C<$NODE> variable contains, the node
216 ID of the node running in the current process. This value is initialised by
217 a call to C<configure>.
218
219 =item $nodeid = node_of $port
220
221 Extracts and returns the node ID from a port ID or a node ID.
222
223 =item configure $profile, key => value...
224
225 =item configure key => value...
226
227 Before a node can talk to other nodes on the network (i.e. enter
228 "distributed mode") it has to configure itself - the minimum a node needs
229 to know is its own name, and optionally it should know the addresses of
230 some other nodes in the network to discover other nodes.
231
232 This function configures a node - it must be called exactly once (or
233 never) before calling other AnyEvent::MP functions.
234
235 The key/value pairs are basically the same ones as documented for the
236 F<aemp> command line utility (sans the set/del prefix), with these additions:
237
238 =over 4
239
240 =item norc => $boolean (default false)
241
242 If true, then the rc file (e.g. F<~/.perl-anyevent-mp>) will I<not>
243 be consulted - all configuraiton options must be specified in the
244 C<configure> call.
245
246 =item force => $boolean (default false)
247
248 IF true, then the values specified in the C<configure> will take
249 precedence over any values configured via the rc file. The default is for
250 the rc file to override any options specified in the program.
251
252 =item secure => $pass->($nodeid)
253
254 In addition to specifying a boolean, you can specify a code reference that
255 is called for every remote execution attempt - the execution request is
256 granted iff the callback returns a true value.
257
258 See F<semp setsecure> for more info.
259
260 =back
261
262 =over 4
263
264 =item step 1, gathering configuration from profiles
265
266 The function first looks up a profile in the aemp configuration (see the
267 L<aemp> commandline utility). The profile name can be specified via the
268 named C<profile> parameter or can simply be the first parameter). If it is
269 missing, then the nodename (F<uname -n>) will be used as profile name.
270
271 The profile data is then gathered as follows:
272
273 First, all remaining key => value pairs (all of which are conveniently
274 undocumented at the moment) will be interpreted as configuration
275 data. Then they will be overwritten by any values specified in the global
276 default configuration (see the F<aemp> utility), then the chain of
277 profiles chosen by the profile name (and any C<parent> attributes).
278
279 That means that the values specified in the profile have highest priority
280 and the values specified directly via C<configure> have lowest priority,
281 and can only be used to specify defaults.
282
283 If the profile specifies a node ID, then this will become the node ID of
284 this process. If not, then the profile name will be used as node ID, with
285 a unique randoms tring (C</%u>) appended.
286
287 The node ID can contain some C<%> sequences that are expanded: C<%n>
288 is expanded to the local nodename, C<%u> is replaced by a random
289 strign to make the node unique. For example, the F<aemp> commandline
290 utility uses C<aemp/%n/%u> as nodename, which might expand to
291 C<aemp/cerebro/ZQDGSIkRhEZQDGSIkRhE>.
292
293 =item step 2, bind listener sockets
294
295 The next step is to look up the binds in the profile, followed by binding
296 aemp protocol listeners on all binds specified (it is possible and valid
297 to have no binds, meaning that the node cannot be contacted form the
298 outside. This means the node cannot talk to other nodes that also have no
299 binds, but it can still talk to all "normal" nodes).
300
301 If the profile does not specify a binds list, then a default of C<*> is
302 used, meaning the node will bind on a dynamically-assigned port on every
303 local IP address it finds.
304
305 =item step 3, connect to seed nodes
306
307 As the last step, the seed ID list from the profile is passed to the
308 L<AnyEvent::MP::Global> module, which will then use it to keep
309 connectivity with at least one node at any point in time.
310
311 =back
312
313 Example: become a distributed node using the local node name as profile.
314 This should be the most common form of invocation for "daemon"-type nodes.
315
316 configure
317
318 Example: become a semi-anonymous node. This form is often used for
319 commandline clients.
320
321 configure nodeid => "myscript/%n/%u";
322
323 Example: configure a node using a profile called seed, which is suitable
324 for a seed node as it binds on all local addresses on a fixed port (4040,
325 customary for aemp).
326
327 # use the aemp commandline utility
328 # aemp profile seed binds '*:4040'
329
330 # then use it
331 configure profile => "seed";
332
333 # or simply use aemp from the shell again:
334 # aemp run profile seed
335
336 # or provide a nicer-to-remember nodeid
337 # aemp run profile seed nodeid "$(hostname)"
338
339 =item $SELF
340
341 Contains the current port id while executing C<rcv> callbacks or C<psub>
342 blocks.
343
344 =item *SELF, SELF, %SELF, @SELF...
345
346 Due to some quirks in how perl exports variables, it is impossible to
347 just export C<$SELF>, all the symbols named C<SELF> are exported by this
348 module, but only C<$SELF> is currently used.
349
350 =item snd $port, type => @data
351
352 =item snd $port, @msg
353
354 Send the given message to the given port, which can identify either a
355 local or a remote port, and must be a port ID.
356
357 While the message can be almost anything, it is highly recommended to
358 use a string as first element (a port ID, or some word that indicates a
359 request type etc.) and to consist if only simple perl values (scalars,
360 arrays, hashes) - if you think you need to pass an object, think again.
361
362 The message data logically becomes read-only after a call to this
363 function: modifying any argument (or values referenced by them) is
364 forbidden, as there can be considerable time between the call to C<snd>
365 and the time the message is actually being serialised - in fact, it might
366 never be copied as within the same process it is simply handed to the
367 receiving port.
368
369 The type of data you can transfer depends on the transport protocol: when
370 JSON is used, then only strings, numbers and arrays and hashes consisting
371 of those are allowed (no objects). When Storable is used, then anything
372 that Storable can serialise and deserialise is allowed, and for the local
373 node, anything can be passed. Best rely only on the common denominator of
374 these.
375
376 =item $local_port = port
377
378 Create a new local port object and returns its port ID. Initially it has
379 no callbacks set and will throw an error when it receives messages.
380
381 =item $local_port = port { my @msg = @_ }
382
383 Creates a new local port, and returns its ID. Semantically the same as
384 creating a port and calling C<rcv $port, $callback> on it.
385
386 The block will be called for every message received on the port, with the
387 global variable C<$SELF> set to the port ID. Runtime errors will cause the
388 port to be C<kil>ed. The message will be passed as-is, no extra argument
389 (i.e. no port ID) will be passed to the callback.
390
391 If you want to stop/destroy the port, simply C<kil> it:
392
393 my $port = port {
394 my @msg = @_;
395 ...
396 kil $SELF;
397 };
398
399 =cut
400
401 sub rcv($@);
402
403 sub _kilme {
404 die "received message on port without callback";
405 }
406
407 sub port(;&) {
408 my $id = $UNIQ . ++$ID;
409 my $port = "$NODE#$id";
410
411 rcv $port, shift || \&_kilme;
412
413 $port
414 }
415
416 =item rcv $local_port, $callback->(@msg)
417
418 Replaces the default callback on the specified port. There is no way to
419 remove the default callback: use C<sub { }> to disable it, or better
420 C<kil> the port when it is no longer needed.
421
422 The global C<$SELF> (exported by this module) contains C<$port> while
423 executing the callback. Runtime errors during callback execution will
424 result in the port being C<kil>ed.
425
426 The default callback received all messages not matched by a more specific
427 C<tag> match.
428
429 =item rcv $local_port, tag => $callback->(@msg_without_tag), ...
430
431 Register (or replace) callbacks to be called on messages starting with the
432 given tag on the given port (and return the port), or unregister it (when
433 C<$callback> is C<$undef> or missing). There can only be one callback
434 registered for each tag.
435
436 The original message will be passed to the callback, after the first
437 element (the tag) has been removed. The callback will use the same
438 environment as the default callback (see above).
439
440 Example: create a port and bind receivers on it in one go.
441
442 my $port = rcv port,
443 msg1 => sub { ... },
444 msg2 => sub { ... },
445 ;
446
447 Example: create a port, bind receivers and send it in a message elsewhere
448 in one go:
449
450 snd $otherport, reply =>
451 rcv port,
452 msg1 => sub { ... },
453 ...
454 ;
455
456 Example: temporarily register a rcv callback for a tag matching some port
457 (e.g. for an rpc reply) and unregister it after a message was received.
458
459 rcv $port, $otherport => sub {
460 my @reply = @_;
461
462 rcv $SELF, $otherport;
463 };
464
465 =cut
466
467 sub rcv($@) {
468 my $port = shift;
469 my ($nodeid, $portid) = split /#/, $port, 2;
470
471 $NODE{$nodeid} == $NODE{""}
472 or Carp::croak "$port: rcv can only be called on local ports, caught";
473
474 while (@_) {
475 if (ref $_[0]) {
476 if (my $self = $PORT_DATA{$portid}) {
477 "AnyEvent::MP::Port" eq ref $self
478 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
479
480 $self->[0] = shift;
481 } else {
482 my $cb = shift;
483 $PORT{$portid} = sub {
484 local $SELF = $port;
485 eval { &$cb }; _self_die if $@;
486 };
487 }
488 } elsif (defined $_[0]) {
489 my $self = $PORT_DATA{$portid} ||= do {
490 my $self = bless [$PORT{$portid} || sub { }, { }, $port], "AnyEvent::MP::Port";
491
492 $PORT{$portid} = sub {
493 local $SELF = $port;
494
495 if (my $cb = $self->[1]{$_[0]}) {
496 shift;
497 eval { &$cb }; _self_die if $@;
498 } else {
499 &{ $self->[0] };
500 }
501 };
502
503 $self
504 };
505
506 "AnyEvent::MP::Port" eq ref $self
507 or Carp::croak "$port: rcv can only be called on message matching ports, caught";
508
509 my ($tag, $cb) = splice @_, 0, 2;
510
511 if (defined $cb) {
512 $self->[1]{$tag} = $cb;
513 } else {
514 delete $self->[1]{$tag};
515 }
516 }
517 }
518
519 $port
520 }
521
522 =item peval $port, $coderef[, @args]
523
524 Evaluates the given C<$codref> within the contetx of C<$port>, that is,
525 when the code throews an exception the C<$port> will be killed.
526
527 Any remaining args will be passed to the callback. Any return values will
528 be returned to the caller.
529
530 This is useful when you temporarily want to execute code in the context of
531 a port.
532
533 Example: create a port and run some initialisation code in it's context.
534
535 my $port = port { ... };
536
537 peval $port, sub {
538 init
539 or die "unable to init";
540 };
541
542 =cut
543
544 sub peval($$) {
545 local $SELF = shift;
546 my $cb = shift;
547
548 if (wantarray) {
549 my @res = eval { &$cb };
550 _self_die if $@;
551 @res
552 } else {
553 my $res = eval { &$cb };
554 _self_die if $@;
555 $res
556 }
557 }
558
559 =item $closure = psub { BLOCK }
560
561 Remembers C<$SELF> and creates a closure out of the BLOCK. When the
562 closure is executed, sets up the environment in the same way as in C<rcv>
563 callbacks, i.e. runtime errors will cause the port to get C<kil>ed.
564
565 The effect is basically as if it returned C<< sub { peval $SELF, sub {
566 BLOCK }, @_ } >>.
567
568 This is useful when you register callbacks from C<rcv> callbacks:
569
570 rcv delayed_reply => sub {
571 my ($delay, @reply) = @_;
572 my $timer = AE::timer $delay, 0, psub {
573 snd @reply, $SELF;
574 };
575 };
576
577 =cut
578
579 sub psub(&) {
580 my $cb = shift;
581
582 my $port = $SELF
583 or Carp::croak "psub can only be called from within rcv or psub callbacks, not";
584
585 sub {
586 local $SELF = $port;
587
588 if (wantarray) {
589 my @res = eval { &$cb };
590 _self_die if $@;
591 @res
592 } else {
593 my $res = eval { &$cb };
594 _self_die if $@;
595 $res
596 }
597 }
598 }
599
600 =item $guard = mon $port, $cb->(@reason) # call $cb when $port dies
601
602 =item $guard = mon $port, $rcvport # kill $rcvport when $port dies
603
604 =item $guard = mon $port # kill $SELF when $port dies
605
606 =item $guard = mon $port, $rcvport, @msg # send a message when $port dies
607
608 Monitor the given port and do something when the port is killed or
609 messages to it were lost, and optionally return a guard that can be used
610 to stop monitoring again.
611
612 In the first form (callback), the callback is simply called with any
613 number of C<@reason> elements (no @reason means that the port was deleted
614 "normally"). Note also that I<< the callback B<must> never die >>, so use
615 C<eval> if unsure.
616
617 In the second form (another port given), the other port (C<$rcvport>)
618 will be C<kil>'ed with C<@reason>, if a @reason was specified, i.e. on
619 "normal" kils nothing happens, while under all other conditions, the other
620 port is killed with the same reason.
621
622 The third form (kill self) is the same as the second form, except that
623 C<$rvport> defaults to C<$SELF>.
624
625 In the last form (message), a message of the form C<@msg, @reason> will be
626 C<snd>.
627
628 Monitoring-actions are one-shot: once messages are lost (and a monitoring
629 alert was raised), they are removed and will not trigger again.
630
631 As a rule of thumb, monitoring requests should always monitor a port from
632 a local port (or callback). The reason is that kill messages might get
633 lost, just like any other message. Another less obvious reason is that
634 even monitoring requests can get lost (for example, when the connection
635 to the other node goes down permanently). When monitoring a port locally
636 these problems do not exist.
637
638 C<mon> effectively guarantees that, in the absence of hardware failures,
639 after starting the monitor, either all messages sent to the port will
640 arrive, or the monitoring action will be invoked after possible message
641 loss has been detected. No messages will be lost "in between" (after
642 the first lost message no further messages will be received by the
643 port). After the monitoring action was invoked, further messages might get
644 delivered again.
645
646 Inter-host-connection timeouts and monitoring depend on the transport
647 used. The only transport currently implemented is TCP, and AnyEvent::MP
648 relies on TCP to detect node-downs (this can take 10-15 minutes on a
649 non-idle connection, and usually around two hours for idle connections).
650
651 This means that monitoring is good for program errors and cleaning up
652 stuff eventually, but they are no replacement for a timeout when you need
653 to ensure some maximum latency.
654
655 Example: call a given callback when C<$port> is killed.
656
657 mon $port, sub { warn "port died because of <@_>\n" };
658
659 Example: kill ourselves when C<$port> is killed abnormally.
660
661 mon $port;
662
663 Example: send us a restart message when another C<$port> is killed.
664
665 mon $port, $self => "restart";
666
667 =cut
668
669 sub mon {
670 my ($nodeid, $port) = split /#/, shift, 2;
671
672 my $node = $NODE{$nodeid} || add_node $nodeid;
673
674 my $cb = @_ ? shift : $SELF || Carp::croak 'mon: called with one argument only, but $SELF not set,';
675
676 unless (ref $cb) {
677 if (@_) {
678 # send a kill info message
679 my (@msg) = ($cb, @_);
680 $cb = sub { snd @msg, @_ };
681 } else {
682 # simply kill other port
683 my $port = $cb;
684 $cb = sub { kil $port, @_ if @_ };
685 }
686 }
687
688 $node->monitor ($port, $cb);
689
690 defined wantarray
691 and ($cb += 0, Guard::guard { $node->unmonitor ($port, $cb) })
692 }
693
694 =item $guard = mon_guard $port, $ref, $ref...
695
696 Monitors the given C<$port> and keeps the passed references. When the port
697 is killed, the references will be freed.
698
699 Optionally returns a guard that will stop the monitoring.
700
701 This function is useful when you create e.g. timers or other watchers and
702 want to free them when the port gets killed (note the use of C<psub>):
703
704 $port->rcv (start => sub {
705 my $timer; $timer = mon_guard $port, AE::timer 1, 1, psub {
706 undef $timer if 0.9 < rand;
707 });
708 });
709
710 =cut
711
712 sub mon_guard {
713 my ($port, @refs) = @_;
714
715 #TODO: mon-less form?
716
717 mon $port, sub { 0 && @refs }
718 }
719
720 =item kil $port[, @reason]
721
722 Kill the specified port with the given C<@reason>.
723
724 If no C<@reason> is specified, then the port is killed "normally" -
725 monitor callback will be invoked, but the kil will not cause linked ports
726 (C<mon $mport, $lport> form) to get killed.
727
728 If a C<@reason> is specified, then linked ports (C<mon $mport, $lport>
729 form) get killed with the same reason.
730
731 Runtime errors while evaluating C<rcv> callbacks or inside C<psub> blocks
732 will be reported as reason C<< die => $@ >>.
733
734 Transport/communication errors are reported as C<< transport_error =>
735 $message >>.
736
737 =cut
738
739 =item $port = spawn $node, $initfunc[, @initdata]
740
741 Creates a port on the node C<$node> (which can also be a port ID, in which
742 case it's the node where that port resides).
743
744 The port ID of the newly created port is returned immediately, and it is
745 possible to immediately start sending messages or to monitor the port.
746
747 After the port has been created, the init function is called on the remote
748 node, in the same context as a C<rcv> callback. This function must be a
749 fully-qualified function name (e.g. C<MyApp::Chat::Server::init>). To
750 specify a function in the main program, use C<::name>.
751
752 If the function doesn't exist, then the node tries to C<require>
753 the package, then the package above the package and so on (e.g.
754 C<MyApp::Chat::Server>, C<MyApp::Chat>, C<MyApp>) until the function
755 exists or it runs out of package names.
756
757 The init function is then called with the newly-created port as context
758 object (C<$SELF>) and the C<@initdata> values as arguments. It I<must>
759 call one of the C<rcv> functions to set callbacks on C<$SELF>, otherwise
760 the port might not get created.
761
762 A common idiom is to pass a local port, immediately monitor the spawned
763 port, and in the remote init function, immediately monitor the passed
764 local port. This two-way monitoring ensures that both ports get cleaned up
765 when there is a problem.
766
767 C<spawn> guarantees that the C<$initfunc> has no visible effects on the
768 caller before C<spawn> returns (by delaying invocation when spawn is
769 called for the local node).
770
771 Example: spawn a chat server port on C<$othernode>.
772
773 # this node, executed from within a port context:
774 my $server = spawn $othernode, "MyApp::Chat::Server::connect", $SELF;
775 mon $server;
776
777 # init function on C<$othernode>
778 sub connect {
779 my ($srcport) = @_;
780
781 mon $srcport;
782
783 rcv $SELF, sub {
784 ...
785 };
786 }
787
788 =cut
789
790 sub _spawn {
791 my $port = shift;
792 my $init = shift;
793
794 # rcv will create the actual port
795 local $SELF = "$NODE#$port";
796 eval {
797 &{ load_func $init }
798 };
799 _self_die if $@;
800 }
801
802 sub spawn(@) {
803 my ($nodeid, undef) = split /#/, shift, 2;
804
805 my $id = $RUNIQ . ++$ID;
806
807 $_[0] =~ /::/
808 or Carp::croak "spawn init function must be a fully-qualified name, caught";
809
810 snd_to_func $nodeid, "AnyEvent::MP::_spawn" => $id, @_;
811
812 "$nodeid#$id"
813 }
814
815
816 =item after $timeout, @msg
817
818 =item after $timeout, $callback
819
820 Either sends the given message, or call the given callback, after the
821 specified number of seconds.
822
823 This is simply a utility function that comes in handy at times - the
824 AnyEvent::MP author is not convinced of the wisdom of having it, though,
825 so it may go away in the future.
826
827 =cut
828
829 sub after($@) {
830 my ($timeout, @action) = @_;
831
832 my $t; $t = AE::timer $timeout, 0, sub {
833 undef $t;
834 ref $action[0]
835 ? $action[0]()
836 : snd @action;
837 };
838 }
839
840 #=item $cb2 = timeout $seconds, $cb[, @args]
841
842 =item cal $port, @msg, $callback[, $timeout]
843
844 A simple form of RPC - sends a message to the given C<$port> with the
845 given contents (C<@msg>), but adds a reply port to the message.
846
847 The reply port is created temporarily just for the purpose of receiving
848 the reply, and will be C<kil>ed when no longer needed.
849
850 A reply message sent to the port is passed to the C<$callback> as-is.
851
852 If an optional time-out (in seconds) is given and it is not C<undef>,
853 then the callback will be called without any arguments after the time-out
854 elapsed and the port is C<kil>ed.
855
856 If no time-out is given (or it is C<undef>), then the local port will
857 monitor the remote port instead, so it eventually gets cleaned-up.
858
859 Currently this function returns the temporary port, but this "feature"
860 might go in future versions unless you can make a convincing case that
861 this is indeed useful for something.
862
863 =cut
864
865 sub cal(@) {
866 my $timeout = ref $_[-1] ? undef : pop;
867 my $cb = pop;
868
869 my $port = port {
870 undef $timeout;
871 kil $SELF;
872 &$cb;
873 };
874
875 if (defined $timeout) {
876 $timeout = AE::timer $timeout, 0, sub {
877 undef $timeout;
878 kil $port;
879 $cb->();
880 };
881 } else {
882 mon $_[0], sub {
883 kil $port;
884 $cb->();
885 };
886 }
887
888 push @_, $port;
889 &snd;
890
891 $port
892 }
893
894 =back
895
896 =head1 DISTRIBUTED DATABASE
897
898 AnyEvent::MP comes with a simple distributed database. The database will
899 be mirrored asynchronously on all global nodes. Other nodes bind to one
900 of the global nodes for their needs. Every node has a "local database"
901 which contains all the values that are set locally. All local databases
902 are merged together to form the global database, which can be queried.
903
904 The database structure is that of a two-level hash - the database hash
905 contains hashes which contain values, similarly to a perl hash of hashes,
906 i.e.:
907
908 $DATABASE{$family}{$subkey} = $value
909
910 The top level hash key is called "family", and the second-level hash key
911 is called "subkey" or simply "key".
912
913 The family must be alphanumeric, i.e. start with a letter and consist
914 of letters, digits, underscores and colons (C<[A-Za-z][A-Za-z0-9_:]*>,
915 pretty much like Perl module names.
916
917 As the family namespace is global, it is recommended to prefix family names
918 with the name of the application or module using it.
919
920 The subkeys must be non-empty strings, with no further restrictions.
921
922 The values should preferably be strings, but other perl scalars should
923 work as well (such as C<undef>, arrays and hashes).
924
925 Every database entry is owned by one node - adding the same family/subkey
926 combination on multiple nodes will not cause discomfort for AnyEvent::MP,
927 but the result might be nondeterministic, i.e. the key might have
928 different values on different nodes.
929
930 Different subkeys in the same family can be owned by different nodes
931 without problems, and in fact, this is the common method to create worker
932 pools. For example, a worker port for image scaling might do this:
933
934 db_set my_image_scalers => $port;
935
936 And clients looking for an image scaler will want to get the
937 C<my_image_scalers> keys from time to time:
938
939 db_keys my_image_scalers => sub {
940 @ports = @{ $_[0] };
941 };
942
943 Or better yet, they want to monitor the database family, so they always
944 have a reasonable up-to-date copy:
945
946 db_mon my_image_scalers => sub {
947 @ports = keys %{ $_[0] };
948 };
949
950 In general, you can set or delete single subkeys, but query and monitor
951 whole families only.
952
953 If you feel the need to monitor or query a single subkey, try giving it
954 it's own family.
955
956 =over
957
958 =item db_set $family => $subkey [=> $value]
959
960 Sets (or replaces) a key to the database - if C<$value> is omitted,
961 C<undef> is used instead.
962
963 =item db_del $family => $subkey...
964
965 Deletes one or more subkeys from the database family.
966
967 =item $guard = db_reg $family => $subkey [=> $value]
968
969 Sets the key on the database and returns a guard. When the guard is
970 destroyed, the key is deleted from the database. If C<$value> is missing,
971 then C<undef> is used.
972
973 =item db_family $family => $cb->(\%familyhash)
974
975 Queries the named database C<$family> and call the callback with the
976 family represented as a hash. You can keep and freely modify the hash.
977
978 =item db_keys $family => $cb->(\@keys)
979
980 Same as C<db_family>, except it only queries the family I<subkeys> and passes
981 them as array reference to the callback.
982
983 =item db_values $family => $cb->(\@values)
984
985 Same as C<db_family>, except it only queries the family I<values> and passes them
986 as array reference to the callback.
987
988 =item $guard = db_mon $family => $cb->($familyhash, \@added, \@changed, \@deleted)
989
990 Creates a monitor on the given database family. Each time a key is set
991 or or is deleted the callback is called with a hash containing the
992 database family and three lists of added, changed and deleted subkeys,
993 respectively. If no keys have changed then the array reference might be
994 C<undef> or even missing.
995
996 The family hash reference and the key arrays belong to AnyEvent::MP and
997 B<must not be modified or stored> by the callback. When in doubt, make a
998 copy.
999
1000 As soon as possible after the monitoring starts, the callback will be
1001 called with the intiial contents of the family, even if it is empty,
1002 i.e. there will always be a timely call to the callback with the current
1003 contents.
1004
1005 It is possible that the callback is called with a change event even though
1006 the subkey is already present and the value has not changed.
1007
1008 The monitoring stops when the guard object is destroyed.
1009
1010 Example: on every change to the family "mygroup", print out all keys.
1011
1012 my $guard = db_mon mygroup => sub {
1013 my ($family, $a, $c, $d) = @_;
1014 print "mygroup members: ", (join " ", keys %$family), "\n";
1015 };
1016
1017 Exmaple: wait until the family "My::Module::workers" is non-empty.
1018
1019 my $guard; $guard = db_mon My::Module::workers => sub {
1020 my ($family, $a, $c, $d) = @_;
1021 return unless %$family;
1022 undef $guard;
1023 print "My::Module::workers now nonempty\n";
1024 };
1025
1026 Example: print all changes to the family "AnyRvent::Fantasy::Module".
1027
1028 my $guard = db_mon AnyRvent::Fantasy::Module => sub {
1029 my ($family, $a, $c, $d) = @_;
1030
1031 print "+$_=$family->{$_}\n" for @$a;
1032 print "*$_=$family->{$_}\n" for @$c;
1033 print "-$_=$family->{$_}\n" for @$d;
1034 };
1035
1036 =cut
1037
1038 =back
1039
1040 =head1 AnyEvent::MP vs. Distributed Erlang
1041
1042 AnyEvent::MP got lots of its ideas from distributed Erlang (Erlang node
1043 == aemp node, Erlang process == aemp port), so many of the documents and
1044 programming techniques employed by Erlang apply to AnyEvent::MP. Here is a
1045 sample:
1046
1047 http://www.erlang.se/doc/programming_rules.shtml
1048 http://erlang.org/doc/getting_started/part_frame.html # chapters 3 and 4
1049 http://erlang.org/download/erlang-book-part1.pdf # chapters 5 and 6
1050 http://erlang.org/download/armstrong_thesis_2003.pdf # chapters 4 and 5
1051
1052 Despite the similarities, there are also some important differences:
1053
1054 =over 4
1055
1056 =item * Node IDs are arbitrary strings in AEMP.
1057
1058 Erlang relies on special naming and DNS to work everywhere in the same
1059 way. AEMP relies on each node somehow knowing its own address(es) (e.g. by
1060 configuration or DNS), and possibly the addresses of some seed nodes, but
1061 will otherwise discover other nodes (and their IDs) itself.
1062
1063 =item * Erlang has a "remote ports are like local ports" philosophy, AEMP
1064 uses "local ports are like remote ports".
1065
1066 The failure modes for local ports are quite different (runtime errors
1067 only) then for remote ports - when a local port dies, you I<know> it dies,
1068 when a connection to another node dies, you know nothing about the other
1069 port.
1070
1071 Erlang pretends remote ports are as reliable as local ports, even when
1072 they are not.
1073
1074 AEMP encourages a "treat remote ports differently" philosophy, with local
1075 ports being the special case/exception, where transport errors cannot
1076 occur.
1077
1078 =item * Erlang uses processes and a mailbox, AEMP does not queue.
1079
1080 Erlang uses processes that selectively receive messages out of order, and
1081 therefore needs a queue. AEMP is event based, queuing messages would serve
1082 no useful purpose. For the same reason the pattern-matching abilities
1083 of AnyEvent::MP are more limited, as there is little need to be able to
1084 filter messages without dequeuing them.
1085
1086 This is not a philosophical difference, but simply stems from AnyEvent::MP
1087 being event-based, while Erlang is process-based.
1088
1089 You cna have a look at L<Coro::MP> for a more Erlang-like process model on
1090 top of AEMP and Coro threads.
1091
1092 =item * Erlang sends are synchronous, AEMP sends are asynchronous.
1093
1094 Sending messages in Erlang is synchronous and blocks the process until
1095 a conenction has been established and the message sent (and so does not
1096 need a queue that can overflow). AEMP sends return immediately, connection
1097 establishment is handled in the background.
1098
1099 =item * Erlang suffers from silent message loss, AEMP does not.
1100
1101 Erlang implements few guarantees on messages delivery - messages can get
1102 lost without any of the processes realising it (i.e. you send messages a,
1103 b, and c, and the other side only receives messages a and c).
1104
1105 AEMP guarantees (modulo hardware errors) correct ordering, and the
1106 guarantee that after one message is lost, all following ones sent to the
1107 same port are lost as well, until monitoring raises an error, so there are
1108 no silent "holes" in the message sequence.
1109
1110 If you want your software to be very reliable, you have to cope with
1111 corrupted and even out-of-order messages in both Erlang and AEMP. AEMP
1112 simply tries to work better in common error cases, such as when a network
1113 link goes down.
1114
1115 =item * Erlang can send messages to the wrong port, AEMP does not.
1116
1117 In Erlang it is quite likely that a node that restarts reuses an Erlang
1118 process ID known to other nodes for a completely different process,
1119 causing messages destined for that process to end up in an unrelated
1120 process.
1121
1122 AEMP does not reuse port IDs, so old messages or old port IDs floating
1123 around in the network will not be sent to an unrelated port.
1124
1125 =item * Erlang uses unprotected connections, AEMP uses secure
1126 authentication and can use TLS.
1127
1128 AEMP can use a proven protocol - TLS - to protect connections and
1129 securely authenticate nodes.
1130
1131 =item * The AEMP protocol is optimised for both text-based and binary
1132 communications.
1133
1134 The AEMP protocol, unlike the Erlang protocol, supports both programming
1135 language independent text-only protocols (good for debugging), and binary,
1136 language-specific serialisers (e.g. Storable). By default, unless TLS is
1137 used, the protocol is actually completely text-based.
1138
1139 It has also been carefully designed to be implementable in other languages
1140 with a minimum of work while gracefully degrading functionality to make the
1141 protocol simple.
1142
1143 =item * AEMP has more flexible monitoring options than Erlang.
1144
1145 In Erlang, you can chose to receive I<all> exit signals as messages or
1146 I<none>, there is no in-between, so monitoring single Erlang processes is
1147 difficult to implement.
1148
1149 Monitoring in AEMP is more flexible than in Erlang, as one can choose
1150 between automatic kill, exit message or callback on a per-port basis.
1151
1152 =item * Erlang tries to hide remote/local connections, AEMP does not.
1153
1154 Monitoring in Erlang is not an indicator of process death/crashes, in the
1155 same way as linking is (except linking is unreliable in Erlang).
1156
1157 In AEMP, you don't "look up" registered port names or send to named ports
1158 that might or might not be persistent. Instead, you normally spawn a port
1159 on the remote node. The init function monitors you, and you monitor the
1160 remote port. Since both monitors are local to the node, they are much more
1161 reliable (no need for C<spawn_link>).
1162
1163 This also saves round-trips and avoids sending messages to the wrong port
1164 (hard to do in Erlang).
1165
1166 =back
1167
1168 =head1 RATIONALE
1169
1170 =over 4
1171
1172 =item Why strings for port and node IDs, why not objects?
1173
1174 We considered "objects", but found that the actual number of methods
1175 that can be called are quite low. Since port and node IDs travel over
1176 the network frequently, the serialising/deserialising would add lots of
1177 overhead, as well as having to keep a proxy object everywhere.
1178
1179 Strings can easily be printed, easily serialised etc. and need no special
1180 procedures to be "valid".
1181
1182 And as a result, a port with just a default receiver consists of a single
1183 code reference stored in a global hash - it can't become much cheaper.
1184
1185 =item Why favour JSON, why not a real serialising format such as Storable?
1186
1187 In fact, any AnyEvent::MP node will happily accept Storable as framing
1188 format, but currently there is no way to make a node use Storable by
1189 default (although all nodes will accept it).
1190
1191 The default framing protocol is JSON because a) JSON::XS is many times
1192 faster for small messages and b) most importantly, after years of
1193 experience we found that object serialisation is causing more problems
1194 than it solves: Just like function calls, objects simply do not travel
1195 easily over the network, mostly because they will always be a copy, so you
1196 always have to re-think your design.
1197
1198 Keeping your messages simple, concentrating on data structures rather than
1199 objects, will keep your messages clean, tidy and efficient.
1200
1201 =back
1202
1203 =head1 SEE ALSO
1204
1205 L<AnyEvent::MP::Intro> - a gentle introduction.
1206
1207 L<AnyEvent::MP::Kernel> - more, lower-level, stuff.
1208
1209 L<AnyEvent::MP::Global> - network maintenance and port groups, to find
1210 your applications.
1211
1212 L<AnyEvent::MP::DataConn> - establish data connections between nodes.
1213
1214 L<AnyEvent::MP::LogCatcher> - simple service to display log messages from
1215 all nodes.
1216
1217 L<AnyEvent>.
1218
1219 =head1 AUTHOR
1220
1221 Marc Lehmann <schmorp@schmorp.de>
1222 http://home.schmorp.de/
1223
1224 =cut
1225
1226 1
1227